A Comparison of Transects and Spot Mapping in Oak-Pine Woodlands of California

Abstract

We compared density estimates of birds using the spot-mapping and transect methods in grazed and ungrazed habitats in foothill woodlands of the western Sierra Nevada, California. As judged by the percentage error of transect estimates in relation to those from spot mapping, the exponential polynomial estimator from program TRANSECT (Laake et al. 1979) consistently gave better estimates of density than either the Fourier series or exponential power series estimators. Distance estimates grouped into bands of some specified width consistently performed better than ungrouped data, although no single band width was best in all cases. Although several alternatives tested in this study gave density estimates with errors only about half those by a variety of previous methods, even the best alternatives had mean errors of about 25% and maximum errors ranging from 37% to 109%. Errors involved both overand underestimation of density in relation to spot-mapping estimates, even for the same species on the same site in different months or by different observers. The two habitats affected TRANSECT estimates of density differently, and observers differed significantly in the accuracy of their estimates. Because of these problems, transect methods could lead to misinterpretation of real abundance patterns by masking differences between years or sites, even to the extent of suggesting that a species was more abundant in a given year or site than another when the reverse was true. As sample sizes increased, both overestimates and underestimates of density by transect methods converged on density estimates by spot mapping, suggesting a real correspondence in density estimation by these two methods. Empirical assessment of the sample size needed to use program TRANSECT indicated that a data set should include at least 100 records per species--far more than the number normally used to estimate densities by transects.